To evaluate the effectiveness and cost-effectiveness of two complementary interventions, using familial breast cancer as a model condition. The primary care intervention consisted of providing computerised referral guidelines and related education to GPs. The nurse counsellor intervention evaluated genetic nurses as substitutes for specialist geneticists in the initial assessment and management of referred patients.
The computerised referral guidelines study was a pragmatic, cluster randomised controlled trial (RCT) with general practices randomised to intervention or control groups. The nurse counsellor intervention was tested in two concurrent RCTs conducted in separate UK health service locations, using predetermined definitions of equivalence.
The computerised referral guidelines trial took place in general practices in Scotland from November 2000 to June 2001. The nurse counsellor intervention took place in a regional genetics clinic in Scotland, and in two health authorities in Wales served by a single genetics service during 2001.
The computerised referral guidelines study involved GPs and referred patients. Both nurse counsellor intervention trials included women referred for the first time, aged 18 years or over and whose main concern was family history of breast cancer.
The software system was developed with GPs, presenting cancer genetic referral guidelines in a checklist approach. Intervention GPs were invited to postgraduate update education sessions, and both intervention and control practices received paper-based guidelines. The intervention period was November 2000 to June 2001. For the nurse counsellor trial, trial 1 ran outpatient sessions with the same appointment length as the standard service offered by geneticists, but the nurse counsellor saw new patients at the first appointment and referred back to the GP or on to a clinical geneticist according to locally developed protocol, under the supervision of a consultant geneticist. The control intervention was the current service, which comprised an initial and a follow-up appointment with a clinical geneticist. In trial 2, a nurse counsellor ran outpatient sessions with the same appointment length as the new consultant-based cancer genetics service and new patients were seen at the first appointment and referred as in trial 1. The control intervention was a new service, and comprised collection of family history by telephone followed by a consultation with a clinical assistant or a specialist registrar, supervised by a consultant. The intervention was implemented between 1998 and 2001.
Main outcome measures
In the software system trial, the primary outcome was GPs' confidence in their management of patients with concerns about family history of breast cancer. For the nurse counsellor trial, the primary outcome was patient anxiety, measured using standard scales.
In the software system trial, 57 practices (230 GPs) were randomised to the intervention group and 29 (116 GPs) to the control group. No statistically significant differences were detected in GPs' confidence or any other outcomes. Fewer than half of the intervention GPs were aware of the software, and only 22 reported using it in practice. The estimated total cost was GBP3.12 per CD-ROM distributed (2001 prices). For the two arms of the nurse counsellor trial, 289 patients (193 intervention, 96 control) and 297 patients (197 intervention and 100 control) consented, were randomised, returned a baseline questionnaire and attended the clinic for trials 1 and 2 respectively. The analysis in both cases suggested equivalence in all anxiety scores, and no statistically significant differences were detected in other outcomes in either trial. A cost-minimisation analysis suggested that the cost per counselling episode was GBP10.23 lower in intervention arm than in the control arm and GBP10.89 higher in the intervention arm than in the control arm (2001 prices) for trials 1 and 2, respectively. Taking the trials together, the costs were sensitive to the grades of doctors and the time spent in consultant supervision of the nurse counsellor, but they were only slightly affected by the grade of nurse counsellor, the selected discount rate and the lifespan of equipment.
Computer-based systems in the primary care intervention cannot be recommended for widespread use without further evaluation and testing in real practice settings. Genetic nurse counsellors may be a cost-effective alternative to assessment by doctors. This trial does not provide definitive evidence that the general policy of employing genetics nurse counsellors is sound, as it was based on only three individuals. Future evaluations of computer-based decision support systems for primary care must first address their efficacy under ideal conditions, identify barriers to the use of such systems in practice, and provide evidence of the impact of the policy of such systems in routine practice. The nurse counsellor trial should be replicated in other settings to provide reassurance of the generalisability of the intervention and other models of nurse-based assessment, such as in outreach clinics, should be developed and evaluated. The design of future evaluations of professional substitution should also address issues such as the effect of different levels of training and experience of nurse counsellors, and learning effects.